ECG) And Electroencephalogram Research Articles

Effects of unconscious tactile stimuli on autonomic nervous activity and afferent signal processing

Autonomic nervous system (ANS) is a mechanism that regulates our internal environment. In recent years, the interest in how tactile stimuli presented directly to the body affect ANS function and cortical processing in humans has been renewed. However, it is not yet clear how subtle tactile stimuli below the level of consciousness affect human heart rate and cortical processing. To examine this, subthreshold electrical stimuli were presented to the left forearm of 43 participants during an image-viewing task, and electrocardiogram (ECG) and electroencephalogram (EEG) data were collected. The changes in the R-wave interval of the ECG immediately after the subthreshold electrical presentation and heartbeat-evoked potential (HEP), the afferent signal processing of cardiac activity, were measured. The results showed that heart rate decelerated immediately after the presentation of subthreshold electrical stimuli. The HEP during stimulus presentation was amplified for participants with greater heart rate acceleration immediately after this deceleration. The magnitude of these effects depended on the type of the subthreshold tactile stimuli. The results suggest that even with subthreshold stimulation, the changes in autonomic activity associated with orienting response and related afferent signal processing differ depending on the clarity of the tactile stimuli.

Effects of Vibroacoustic Stimulation on Psychological, Physiological, and Cognitive Stress.

Global stress is widespread in today's post-pandemic world of political and economic uncertainty. Vibroacoustic technology is a vibrotactile intervention with multiple uses, but its impact on stress lacks interpretation. This research assessed if the vibroacoustic technology of a Vibroacoustic Sound Massage (VSM) can reduce psychological, physiological, and cognitive stress. The Perceived Stress Scale (PSS-10) and electrocardiogram (ECG) and electroencephalogram (EEG) biosignals were used to quantify results. Participants were divided into Low-Stress and High-Stress groups. The ECG results show VSM increased parasympathetic activity for all participants, with the Low-Stress group being more affected. The EEG results indicate increased concentration, reduced arousal, and increased relaxation, with participant well-being non-significantly affected, though variability in this metric was homogenised after VSM. Together, these results validate VSM as an effective support tool for stress management; however, further research is required.

Utilizing fast Fourier transform in the processing of biomedical signals: An analytical approach

Fast Fourier Transform (FFT) is an indispensable tool in biomedical engineering, which optimizes the computational process of the traditional Discrete Fourier Transform (DFT) and effectively reduces the computational complexity. This paper first introduces the basic principle of FFT and its importance in biomedical signal processing. It focuses on the application of FFT in the analysis of electrocardiogram (ECG) and electroencephalogram (EEG) signals, such as the diagnosis of arrhythmia and the analysis of sleep quality, as well as the importance of FFT in removing biomedical signal noise. In addition, the paper explores FFT discussing its challenges and future directions in biomedical engineering, including the development of new algorithms and integration with machine learning techniques. Finally, the article summarizes the applications of FFTs in biomedical engineering and their importance. It emphasizes the need for continued research and development of FFTs and their related techniques in the biomedical field.

Potential preventative impact of aloe-emodin nanoparticles on cerebral stroke-associated myocardial injury by targeting myeloperoxidase: In supporting with In silico and In vivo studies

The present study examined the potential neuroprotective effects of aloe-emodin (AE) nanoparticles on the cerebral stroke-associated target protein myeloperoxidase (MPO). We investigated the binding interactions between AE and MPO through molecular docking and molecular dynamics simulations. Molecular docking results indicated that AE exhibited a binding energy of −6.9 kcal/mol, whereas it was −7.7 kcal/mol for 2-{[3,5-bis(trifluoromethyl)benzyl]amino}-n-hydroxy-6-oxo-1,6-dihydropyrimidine-5-carboxamide (CCl). Furthermore, molecular dynamics studies demonstrated that AE possesses a stronger binding affinity (−57.137 ± 13.198 kJ/mol) than does CCl (−22.793 ± 30.727 kJ/mol), suggesting that AE has a more substantial inhibitory effect on MPO than does CCl. Despite the therapeutic potential of AE for neurodegenerative disorders, its bioavailability is limited within the body. A proposed hypothesis to enhance the bioavailability of AE is its conversion into aloe-emodin nanoparticles (AENP). The AENPs synthesized through a fabrication method were spherical with a consistent diameter of 104.4 ± 7.9 nm and a polydispersity index ranging from 0.525 to 0.586. In rats experiencing cerebral stroke, there was a notable increase in cerebral infarction size; abnormalities in electrocardiogram (ECG) and electroencephalogram (EEG) patterns; a decrease in brain and cardiac antioxidant activities; and an increase in myeloperoxidase levels compared to those in normal rats. Compared with AE treatment, AENP treatment significantly ameliorated cerebral infarction, normalized ECG and EEG patterns, enhanced brain and cardiac antioxidant activities, and reduced MPO levels in stroke rats. Histopathological evaluations revealed pronounced alterations in the rat hippocampus, with pyknotic nuclei, disarray and loosely packed cells, deterioration of cardiac muscle fibers, and extensive damage to cardiac myocytes, in contrast to those in normal rats. AENP treatment mitigated these pathological changes more effectively than AE treatment in both brain and cardiac cells. These findings support that AENP provides considerable protection against stroke-associated myocardial infarction.

Are Subjective Benefits of Mindfulness-Based Stress Reduction Related to Changes in Cardiac and Cortical Responses to a Stress Task?

ObjectivesMindfulness-Based Stress Reduction (MBSR) has been shown to decrease general feelings of distress. It is not known to what extent such beneficial effects are associated with attenuation of physiological responses to challenging psychological situations. The aim of the present study was to examine potential associations between general decreases in subjective distress and changes in acute cardiac and cortical responses during recall of a recent stressful episode between pre- and post-MBSR.MethodFifty-eight MBSR participants (77.6% female, mean age 43.8 years, SD = 13.1) took part in a laboratory examination before and after MBSR. Measurements of the electrocardiogram (ECG) and electroencephalogram (EEG) were performed during the whole protocol including the stress recall task.ResultsThe MBSR group showed overall decreases in general negative affect and an increase in heart rate variability from pre- to post-intervention. Out of six physiological outcomes (heart rate, heart rate variability, theta, alpha, beta, and gamma power at Fz), only a decreased response of gamma power during the stress task was significant. The subjective and physiological changes were unrelated.ConclusionsThe present results suggest that the favourable effects of MBSR on general stress reduction may not be clearly reflected in cortical electrical wave activity or vagal cardiac response during recall of a recent stressful episode. The few changes found in cardiac and cortical activity may be due to either effects independent of general subjective feelings or independent of the intervention altogether.PreregistrationThe study is not preregistered.

Performance Analysis of Vedic Multiplier in Hardware Implementation of Biomedical Applications

As technology advances at a rapid pace, there is an increasing need for real-time digital signal processing (DSP) applications that are efficient and swift. DSPs, or digital signal processors, are crucial components of several engineering disciplines. For processes like convolution and Fourier transforms in DSPs, rapid multiplication is essential. Multiplication is one of the basic mathematical operations used by all applications. Many different multiplier designs have been developed to boost their speed. When compared to array and booth multipliers—the products of decades of hard research—vedic multipliers are among the fastest and lowest power multipliers. The sixteen sutras, or algorithms, that the Vedic Multiplier uses are primarily logical procedures. They are the fastest and most effective because several of them have been proposed using the Urdhava Tiryakbhyam sutra. The purpose of this study is to provide an overview of the numerous biomedical applications of Vedic Multiplier in the wide field of digital signal processing, including denoising of Electrocardiogram (ECG) and Electroencephalogram ( EEG) signal. Particular attention is paid to how current Vedic Multiplier designs have been altered to increase speed and performance metrics.

Light weight neural network for ECG and EEG anomaly detection in IOT edge sensors

Lightweight neural network designed for detecting anomalies in Electrocardiogram (ECG) and Electroencephalogram (EEG) signals at IoT edge sensors. By optimizing neural network architectures, we achieve high accuracy in anomaly detection while minimizing computational demands and memory usage. Experimental results validate the effectiveness of our approach in real-world scenarios, promising improved healthcare monitoring with early detection of abnormal ECG and EEG patterns at the edge.

Solid Reservoir Reference Electrode

Potentiometric sensing requires a stable sensing electrode that is selectively responsive to an analyte, and a reference electrode (RE) that provides a stable and well-defined reference potential in a liquid electrolytic environment. In sensor applications where compactness and ease of integration are essential characteristics, pseudo-REs in the form of doped polymer semiconductors, bare metal films, or a bare chlorinated silver film are typically used. Bulk glass electrodes, such as the Ag/AgCl electrode, are incompatible with sensor integration.We report here a solid reservoir reference electrode (SRRE) that achieves the performance of traditional immersed wire Ag/AgCl REs in a compact, layered structure. The SRRE is a solid, layered analogue of the Ag/AgCl RE, with a chlorinated silver film as metallic electrode, a solid KCl layer serving as a saturated reservoir of chloride, and a porous polydimethylsiloxane (PDMS) layer acting as membrane for exchange of water and ions. We demonstrate two different method to fabricate the porous PDMS in figure A and B. The characteristics of the SRRE, including impedance, noise, and long-term drift, can be tuned by adjustment of the PDMS membrane permeance and KCl solid reservoir size.We have demonstrated temporal stability of the RE open circuit potential (OCP), with drift less than 0.37 mV over 17 hours in deionized water. When compared against bare chlorinated silver, the SRRE exhibits superior stability of OCP versus changes in electrolyte ion-concentration, including solutions of potassium chloride, sodium chloride and pH buffers. The ease of fabrication enables SRRE integration with graphene ion-sensitive field effect transistors (ISFETs) to achieve entirely solid-contact micro ion sensors. The solid-contact micro ion sensor can simultaneously measure pH and Na+, in volumes as low as 20 μL. Using the same layered structure, we further demonstrated fabricated electrocardiogram (ECG) and electroencephalogram (EEG) electrodes with significantly lower impedance and noise when compared to disposable gel electrodes. Finally, we have shown that the SRRE can function in non-aqueous solventssuch as acetonitrile, demonstrating the wide range of SRRE applications. Figure 1

Continuous Biopotential Monitoring via Carbon Nanotubes Paper Composites (CPC) for Sustainable Health Analysis.

Skin-based wearable devices have gained significant attention due to advancements in soft materials and thin-film technologies. Nevertheless, traditional wearable electronics often entail expensive and intricate manufacturing processes and rely on metal-based substrates that are susceptible to corrosion and lack flexibility. In response to these challenges, this paper has emerged with an alternative substrate for wearable electrodes due to its cost-effectiveness and scalability in manufacturing. Paper-based electrodes offer an attractive solution with their inherent properties of high breathability, flexibility, biocompatibility, and tunability. In this study, we introduce carbon nanotube-based paper composites (CPC) electrodes designed for the continuous detection of biopotential signals, such as electrooculography (EOG), electrocardiogram (ECG), and electroencephalogram (EEG). To prevent direct skin contact with carbon nanotubes, we apply various packaging materials, including polydimethylsiloxane (PDMS), Eco-flex, polyimide (PI), and polyurethane (PU). We conduct a comparative analysis of their signal-to-noise ratios in comparison to conventional gel electrodes. Our system demonstrates real-time biopotential monitoring for continuous health tracking, utilizing CPC in conjunction with a portable data acquisition system. The collected data are analyzed to provide accurate heart rates, respiratory rates, and heart rate variability metrics. Additionally, we explore the feasibility using CPC for sleep monitoring by collecting EEG signals.

AS3-SAE: Automatic Sleep Stages Scoring Using Stacked Autoencoders

Purpose: Sleep is a subconscious state, and the brain is active during it. Automatic classification of sleep stages can help identify various diseases. In recent years, automatic sleep monitoring using deep learning networks has attracted the attention of researchers.
 Materials and Methods: In this paper, a deep learning type neural network called Stacked Autoencoders (SAEs) is used for automatically classifying sleep stages. SAEs are a kind of neural network with encoder and decoder blocks. The function of these networks is similar to the human brain and is capable of automatically processing signals; also SAEs are robust to noise. To prove the efficiency of this network, in addition to examining the effect of various biological signals such as Electrocardiogram (ECG) and Electroencephalogram (EEG) on the performance of sleep stage classification, Sleep Heart Health Study (SHHS) and ISRUC standard databases have been used, which include night recordings of 30 and 10 healthy humans, respectively. 
 Results: The accuracy of classifying 2 to 6 classes by SHHS database are 0.995, 0.983, 0.9780, 0.9688, 0.961, and on ISRUC database accuracies are 0.996, 0.994, 0.9511, and 0.9431. Moreover, the proposed network can classify wake, deep sleep, and light sleep using the ECG signal (acc = 0.75, kappa = 0.69).
 Conclusion: In the review of the results, it is concluded that sleep stages classification based on EEG signal has better results, still acquisition of ECG signal and its acceptable results can be a good alternative to use. In addition to its high ability of the proposed method to detect sleep stages, this network is robust to noise, which is very necessary and important for the clinical processing of sleep signals.

Cardiac AC8 Over-Expression Increases Locomotion by Altering Heart-Brain Communication

BackgroundThe central nervous system’s influence on cardiac function is well described; however, direct evidence for signaling from heart to brain remains sparse. Mice with cardiac-selective overexpression of adenylyl cyclase type 8 (TGAC8) display elevated heart rate/contractility and altered neuroautonomic surveillance. ObjectivesIn this study the authors tested whether elevated adenylyl cyclase type 8–dependent signaling at the cardiac cell level affects brain activity and behavior. MethodsA telemetry system was used to record electrocardiogram (ECG) and electroencephalogram (EEG) in TGAC8 and wild-type mice simultaneously. The Granger causality statistical approach evaluated variations in the ECG/EEG relationship. Mouse behavior was assessed via elevated plus maze, open field, light-dark box, and fear conditioning tests. Transcriptomic and proteomic analyses were performed on brain tissue lysates. ResultsBehavioral testing revealed increased locomotor activity in TGAC8 that included a greater total distance traveled (+43%; P < 0.01), a higher average speed (+38%; P < 0.01), and a reduced freezing time (–45%; P < 0.01). Dual-lead telemetry recording confirmed a persistent heart rate elevation with a corresponding reduction in ECG-R-waves interval variability and revealed increased EEG-gamma activity in TGAC8 vs wild-type. Bioinformatic assessment of hippocampal tissue indicated upregulation of dopamine 5, gamma-aminobutyric acid A, and metabotropic glutamate 1/5 receptors, major players in gamma activity generation. Granger causality analyses of ECG and EEG recordings showed a marked increase in informational flow between the TGAC8 heart and brain. ConclusionsPerturbed signals arising from the heart cause changes in brain activity, altering mouse behavior. More specifically, the brain interprets augmented myocardial humoral/functional output as a “sustained exercise-like” situation and responds by activating central nervous system output controlling locomotion.

Bias Analysis in Healthcare Time Series (BAHT) Decision Support Systems from Meta Data.

One of the hindrances in the widespread acceptance of deep learning-based decision support systems in healthcare is bias. Bias in its many forms occurs in the datasets used to train and test deep learning models and is amplified when deployed in the real world, leading to challenges such as model drift. Recent advancements in the field of deep learning have led to the deployment of deployable automated healthcare diagnosis decision support systems at hospitals as well as tele-medicine through IoT devices. Research has been focused primarily on the development and improvement of these systems leaving a gap in the analysis of the fairness. The domain of FAccT ML (fairness, accountability, and transparency) accounts for the analysis of these deployable machine learning systems. In this work, we present a framework for bias analysis in healthcare time series (BAHT) signals such as electrocardiogram (ECG) and electroencephalogram (EEG). BAHT provides a graphical interpretive analysis of bias in the training, testing datasets in terms of protected variables, and analysis of bias amplification by the trained supervised learning model for time series healthcare decision support systems. We thoroughly investigate three prominent time series ECG and EEG healthcare datasets used for model training and research. We show the extensive presence of bias in the datasets leads to potentially biased or unfair machine-learning models. Our experiments also demonstrate the amplification of identified bias with an observed maximum of 66.66%. We investigate the effect of model drift due to unanalyzed bias in datasets and algorithms. Bias mitigation though prudent is a nascent area of research. We present experiments and analyze the most prevalently accepted bias mitigation strategies of under-sampling, oversampling, and the use of synthetic data for balancing the dataset through augmentation. It is important that healthcare models, datasets, and bias mitigation strategies should be properly analyzed for a fair unbiased delivery of service.

Free-Space Optical Communication with an Optimized Lipschitz Exponent for Biosignal Telemetry

Abstract Healthcare monitoring is a rapidly developing network in the field of advanced medical treatment. The network combines the ideology of wireless communication, signal processing, medical information and real-time processing units to support the medical monitoring system. The proposed work focuses on the development of a Free-Space Optical (FSO) system to transmit the biosignals from a remote distance to the physician. Generally, the data transmitted over the FSO system is affected by various atmospheric conditions such as air medium, O2, and H2O molecules. To tackle these problems, the Biosignals Electrocardiogram (ECG) and Electroencephalogram (EEG) are processed in the Optimized Lipschitz Exponent (OLE) function before transmission over the FSO medium. In this novel technique, the OLE function measures the informative data from the biosignals by calculating the local regularities and singularity. This collects the most informative signals and transmits them in the signal over the FSO medium. This particular hybridization helps to transmit the required data without distortion. The Bit Error Rate (BER) of 10−9 is obtained, which satisfies the healthcare monitoring condition. The result section shows that the proposed model has minimum losses compared to the original signal.

Study of Interactive Variation Between Brain and Heart Signals While Listening to the Holy Quran by Fusion Technique

Background: In recent years, much attention has been paid to the impact of spirituality on people’s health. Some signals can alter brain function and affects the autonomic nervous system to reduce blood pressure, heart rate, and anxiety levels. Objectives: This study aimed to investigate the effect of listening to the Holy Quran on the electrocardiogram (ECG) and electroencephalogram (EEG) signals of healthy people with the fusion technique. Materials &amp; Methods: Cardiac signal recording and two brain signal channels in the C3 and C4 areas of 25 female students between 20 and 23 years old were performed in three stages: silence, listening to the Holy Quran, and silence again. We used standard complementary plots, then we matched the circles with different radii (0.1 to 1) on the complementary diagram and extracted the number of intersection points with the hypothetical lines of the complementary plot as a feature. We then examined all possible modes with the support vector machine classifier. A new data fusion technique was used to study the interactions between the heart and the brain. Results: The best accuracy of 98.75% was obtained for a distinction between pre and no-voice using the brain signal. Conclusion: The results of the present study show the effect of listening to the Holy Quran on physiological signals with the fusion technique.

Efficient Clustering-Based electrocardiographic biometric identification

The correct identification of individuals through different biometric traits is becoming increasingly important. Apart from traditional biomarkers (like fingerprints), many alternative measures have been proposed during the last two decades: electrocardiogram (ECG) and electroencephalogram (EEG) signals, iris or facial recognition, conductual traits, etc. Several works have shown that ECG-based recognition is a feasible alternative, either for stand-alone or multi-biometric recognition systems. In this paper, we propose a novel framework for ECG-based biometric identification, consisting of a simple and robust feature extraction approach and a clustering-based feature reduction method, that enables for an efficient and scalable biometric identification. The proposed feature reduction approach is a two phase method: it uses a clustering algorithm to group features according to their similarities first, and then clusters are represented in terms of a prototype vector and associated to the available subjects. On its side, the proposed time-domain feature extraction method is a semi-fiducial procedure, where the well-known Pan–Tompkins algorithm is first used to detect the R wave peaks of the QRS complexes, and then fixed-width time segments are selected for further dimensionality reduction and feature extraction. The resulting combined methods are efficient, robust, scalable and attain excellent results (with up-to 98.6% sensitivity) on all the subjects of the Physikalisch-Technische Bundesanstalt (PTB) database, regardless of their pathological or healthy status. Additionally, we also show how the existing Auto Correlation/Discrete Cosine Transform (AC/DCT)-based non-fiducial feature extraction method can be integrated within our framework, allowing us to attain up to 90.6% sensitivity on the Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database. Since this database is much noisier and has a much lower sampling rate (360 Hz instead of 1000 Hz), we claim that this is a very good result.

A NOVEL ECG AND EEG CLASSIFICATION SYSTEM BASED ON NONLINEAR STATISTICAL FEATURES

Accurate classification of the medical signals is urgently needed in clinical medicine. This paper aims to create a classifier to shorten the time of the classification and ensure the sorting accuracy, which assists physicians in saving diagnostic time and formulating the treatment plans. We create the classifier based on Kolmogorov complexity, Shannon entropy, Higuchi’s Hurst exponent and multifractal features. We obtain a feature value from Kolmogorov complexity, Shannon entropy and Higuchi’s Hurst exponent, and three feature values based on multifractal features to compose a vector and analyze it. Furthermore, we study a vector composed of six multifractal features as a control group. Electrocardiogram (ECG) and electroencephalogram (EEG) signals are applied to examine the performance of the classifier by support vector machine (SVM). The accuracy of ECG signals based on mixed classification (MC–ECG–SVM) reaches 94.17%, which is approximately 15% higher than that of ECG signals only based on multifractal features classification (UC–ECG–SVM). The sensitivities of MC–ECG–SVM and UC–ECG–SVM are 86.09% and 64.54%, respectively. The specificities of MC–ECG–SVM and UC–ECG–SVM are 98.26% and 93.65%, respectively. Analogously, the accuracy, sensitivity, and specificity of EEG signals based on mixed classification (MC–EEG–SVM) reach 95.29%, 96.28%, and 94.55%, respectively. The accuracy, sensitivity, and specificity of EEG signals based on multifractal features classification (UC–EEG–SVM) are 87.40%, 89.28%, and 88.11%, respectively. Therefore, the mixed classification method is more accurate than the classification method only based on multifractal features.

AI for the detection of neurological condition: Parkinson's disease &amp; emotions

Artificial Intelligence (AI) is widely applied by many researchers in the measurement and analysis of signals and images in clinical medicine and the biological sciences. The role of machine learning in processing biomedical signals and its applications in medicine and healthcare is huge, and it is now in a very advanced stage. Several types of biomedical signals have been analyzed by using Deep Learning (DL), Neural Networks (NN), and Artificial Intelligence on Electrocardiogram (ECG) and Electroencephalogram (EEG) signals by many researchers. Parkinson's disease (PD) is a neurodegenerative disorder that progresses over time and is characterized by rigidity, tremor, postural instability, and non-motor symptoms caused by the loss of dopaminergic neurons in the substantia nigra. This paper analyses the current state of the art of EEG analysis using AI techniques for Parkinson's disease detection and emotion detection.

Automated Emotion Identification Using Fourier-Bessel Domain-Based Entropies.

Human dependence on computers is increasing day by day; thus, human interaction with computers must be more dynamic and contextual rather than static or generalized. The development of such devices requires knowledge of the emotional state of the user interacting with it; for this purpose, an emotion recognition system is required. Physiological signals, specifically, electrocardiogram (ECG) and electroencephalogram (EEG), were studied here for the purpose of emotion recognition. This paper proposes novel entropy-based features in the Fourier–Bessel domain instead of the Fourier domain, where frequency resolution is twice that of the latter. Further, to represent such non-stationary signals, the Fourier–Bessel series expansion (FBSE) is used, which has non-stationary basis functions, making it more suitable than the Fourier representation. EEG and ECG signals are decomposed into narrow-band modes using FBSE-based empirical wavelet transform (FBSE-EWT). The proposed entropies of each mode are computed to form the feature vector, which are further used to develop machine learning models. The proposed emotion detection algorithm is evaluated using publicly available DREAMER dataset. K-nearest neighbors (KNN) classifier provides accuracies of 97.84%, 97.91%, and 97.86% for arousal, valence, and dominance classes, respectively. Finally, this paper concludes that the obtained entropy features are suitable for emotion recognition from given physiological signals.

Predicting intraoperative hypotension using deep learning with waveforms of arterial blood pressure, electroencephalogram, and electrocardiogram: Retrospective study.

To develop deep learning models for predicting Interoperative hypotension (IOH) using waveforms from arterial blood pressure (ABP), electrocardiogram (ECG), and electroencephalogram (EEG), and to determine whether combination ABP with EEG or CG improves model performance. Data were retrieved from VitalDB, a public data repository of vital signs taken during surgeries in 10 operating rooms at Seoul National University Hospital from January 6, 2005, to March 1, 2014. Retrospective data from 14,140 adult patients undergoing non-cardiac surgery with general anaesthesia were used. The predictive performances of models trained with different combinations of waveforms were evaluated and compared at time points at 3, 5, 10, 15 minutes before the event. The performance was calculated by area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), sensitivity and specificity. The model performance was better in the model using both ABP and EEG waveforms than in all other models at all time points (3, 5, 10, and 15 minutes before an event) Using high-fidelity ABP and EEG waveforms, the model predicted IOH with a AUROC and AUPRC of 0.935 [0.932 to 0.938] and 0.882 [0.876 to 0.887] at 5 minutes before an IOH event. The output of both ABP and EEG was more calibrated than that using other combinations or ABP alone. The results demonstrate that a predictive deep neural network can be trained using ABP, ECG, and EEG waveforms, and the combination of ABP and EEG improves model performance and calibration.

Detection of Episodes of Sleep Apnea and Hypopnea in ECG and EEG Signals by Machine Learning

The article is devoted to the application of machine learning methods for computerized detection of sleep apnea episodes based on the analysis of single-channel signals of the electrocardiogram (ECG) and electroencephalogram (EEG). To study the possibilities of machine learning to detect apnea based on ECG and EEG analysis, we used Apnea-ECG database and MIT-BIH polysomnographic database from PhysioNet, which contain annotations to each minute of records indicating the presence or absence of apnea/hypopnea at the current time. In order to apply machine learning methods to the problem of automated detection of sleep apnea/hypopnea episodes in ECG and EEG signals, long-term polysomnograms available in MIT-BIH polysomnographic database were segmented according to annotations into shorter sections lasting 30 seconds each. The study used 267 segments lasting 30 seconds for the class "norm", 258 segments for the class "apnea" and 273 segments for the class "hypopnea", a total of 798 simultaneous ECG and EEG recordings. The aim of this work is to identify and compare informative signs of sleep apnea episodes in terms of heart rate variability (HRV) and brain electrical activity, as well as the choice of classification methods that provide the highest accuracy for this task. Features of cardiorhythmograms in time and frequency domains, spectral-temporal and wavelet characteristics, as well as parameters of EEG signals based on energy ratio of EEG rhythms, Hearst index, Higuchi fractal dimension and sample entropy for EEG signals are considered. Using different sets of features, the accuracy of classifiers based on decision trees, discriminant analysis, support vector machines, k-nearest neighbor method, and ensemble training was determined. Based on this, combination of features and classifiers is proposed, which provides the highest accuracy of recognition of sleep apnea episodes according to single-channel ECG and EEG signals, taken separately and in the case of a combination of their features. The best results of classification of signals "norm", "apnea" and "hypopnea" were obtained for the model trained using weighted method k nearest neighbors with 25 features of HRV: the total percentage of correctly identified cases for three classes was 99.9% (797 correctly identified cases of 798). By reducing the number of HRV parameters to 9, the best machine learning result was achieved using the bagging ensemble algorithm with 30 decision trees: the total percentage of correctly identified cases for all three classes was 99.4% (793 correctly identified cases from 798: for "norm" - 265 cases from 267, for "apnea" - 257 cases from 258, for "hypopnea" - 271 cases from 273). The use of EEG parameters as features for apnea/hypopnea recognition showed worse results compared to HRV parameters. In this case, the best result of machine learning was achieved using support vector machines with quadratic kernel function: the total percentage of correctly identified cases for three classes was 91.9% and the signals corresponding to norm were most badly recognized (27 cases were classified as hypopnea, and in 9 cases - as sleep apnea). The combination of HRV and EEG parameters gave the best accuracy of 99.1%, but the results are comparable to using only HRV parameters. The obtained results indicate that HRV parameters allow recognizing sleep apnea and hypopnea with higher accuracy than EEG parameters, but EEG signal undoubtedly reflects signs of sleep apnea/hypopnea and also can be used for apnea recognition.